I have made some measurements on 4 scanners (Canon FS-4000, Polaroid
SS4000, Nikon LS4000, & Minolta Diamage Dual) using the Q60 test image.
After reading Tony's discussion of the Q60, I had assumed the steps were
linear from 1 to 22. They are not. Steps 21 and 22 have half the step
change (2% insstead of 4%). The following is taken from Kodak's Q60
specifications.
The steps (assuming a perfect slide) are:
Q60 Steps Theoretical (%)
0 100
1 87
2 83
3 79
4 75
5 71
6 67
7 63
8 59
9 55
10 51
11 47
12 43
13 39
14 35
15 31
16 27
17 23
18 19
19 15
20 11
21 9
22 7
23 0.0
I took the Q60 images for the Canon FS-4000, Polaroid SS4000, Nikon LS4000,
& Minolta Diamage Dual from , and measured
gray intensities in Photoshop with Kodak's DigitalColor Meter software (Mac
OS 9.04). The Minolta Diamage Dual was included for comparison since this
my current scanner. After adjusting the brightness setting in Vuescan
(ranging 1.4 to 1.6), the Polaroid SS4000, Nikon LS4000 had nearly perfect
linearity down to step 23. The Canon and Minolta had substantial
nonlinearity at step 22.
I've also noticed that a few steps are a little off on all 4 scanners at
the same place, which leads me to beleve these steps on the Q60 are off.
I'm planning on buying a Q60 and making a Kodachrome slide from it to test
the Kodachrome - Scanner response Most of my slides are Kodachrome 64. I
wonder which target would be the best (print or slide)? I had assumed the
slide would be best due to the assumed higher black density.
I also measured the standard deviation of step 23 (rms noise) and got this
result: Minolta= 7.51, Polaroid= 1.78, Canon= 1.36, Nikon= 0.45, units are
least significant bit (lsb) for 8-bit images. My measurement software (NIH
Image) only handles grayscale images. All images were brightness adjusted
for linearity. I'm amazed that the LS4000's noise is substantially lower
than the other scanners. However, the Nikon's noise measurement may be
inaccurate since it's noise is significantly < 1-lsb. I really need a
16-bit image file to accurately measure the Nikon's noise.
My interest in buying a new scanner is that the Minolta does a awful job
scanning Kodachrome (noise, and green and red ghosts in high contrast
scenes (offset by 6 and 15 pixels to the right, respectively, for landscape
slides). I tried 2 Dimage Duals and both had the same ghosts.
Does anyone know of a target with lower gray level steps (eg. 5% and 3%)
since the Polaroid & Nikon scanners can't be fully tested with the Q60?
Also, does anyone have Q60 or similar gray step scans from Nikon's LS-40
and LS-4000 they could send me? I'd like to know if the LS-4000's shadow
detail and noise is much better than the LS-40's.
